public function combineCommonUnits($results, $units_set, $urls_count)
{
$lemmatizator = new Lemmatizator();
$combined_result = [];
$entries_sums = [];
//is neccessary for bodies counting
$zones = $this->allZones();
$queries = array_keys($results);
$counter = new Counter();
foreach ($zones as $zone) {
foreach ($units_set as $unit) {
foreach ($queries as $query) {
if (isset($results[$query][$zone][$unit])) {
$combined_result[$zone][$unit]['entries'][] = $results[$query][$zone][$unit];
//count sums for bodies counting
if ($zone != 'title') {
//titles are not included in bodies counting
foreach ($results[$query][$zone][$unit] as $url_num => $entry_count) {
$entries_sums[$unit][$query][$url_num] = isset($entries_sums[$unit][$query][$url_num]) ? $entries_sums[$unit][$query][$url_num] + $entry_count : $entry_count;
}
}
//get the averages array only for those sets which has been realy counted for entries
$combined_result[$zone][$unit]['average'][] = $counter->getAverage($results[$query][$zone][$unit]);
} else {
//make the false sets of entries for those which have not been realy counted for entries
$combined_result[$zone][$unit]['entries'][] = array_fill(0, $urls_count[$query], 0);
}
}
}
}
// correct the extra FORMs
//========================
//get the list of 'asis' units from units list
//use the fact that we need those which have not no FORMs form right after the asis form
$asis_to_correct_set = [];
foreach ($units_set as $key => $unit) {
if (strpos($unit, 'FORMS') === false && (!isset($units_set[$key + 1]) || strpos($units_set[$key + 1], 'FORMS') === false)) {
$asis_to_correct_set[] = $unit;
}
}
//get the list of FORMs units that should be corrected
//to do this proceed each from asis list with makeUnitNormal function
$forms_to_correct_set = [];
foreach ($asis_to_correct_set as $asis) {
$asis_normal = $lemmatizator->makeUnitNormal($asis);
$asis_normal .= " - FORMS";
$forms_to_correct_set[$asis_normal][] = $asis;
}
//make the array FORMs-unit => all it's asis units
foreach (array_keys($forms_to_correct_set) as $asis_normal) {
$more_key = array_search($asis_normal, $units_set);
array_unshift($forms_to_correct_set[$asis_normal], $units_set[$more_key - 1]);
}
//prepare factors and divider data for each FORMs-unit (with zone)
$correct_factors = [];
foreach (array_keys($forms_to_correct_set) as $form) {
foreach ($zones as $zone) {
$correct_factors[$zone][$form]['divider'] = count($combined_result[$zone][$form]['average']);
foreach ($forms_to_correct_set[$form] as $asis) {
$correct_factors[$zone][$form]['factors'][$asis] = $correct_factors[$zone][$form]['divider'] - count($combined_result[$zone][$asis]['average']);
}
}
}
//not the end of correctness procedure -> to be continued
//=======================================================
//start the combine averages procedure
foreach ($zones as $zone) {
foreach ($units_set as $unit) {
$entries_flattened = [];
foreach ($combined_result[$zone][$unit]['entries'] as $arr) {
$entries_flattened = array_merge($entries_flattened, $arr);
}
// file_put_contents(storage_path("tmp/results.txt"), print_r($combined_result['plain']['КОШКА - FORMS'], true),FILE_APPEND);//FOR TESTING !!!
$combined_result[$zone][$unit]['average'] = $counter->getAverage($combined_result[$zone][$unit]['average']);
$combined_result[$zone][$unit]['entries'] = $entries_flattened;
}
}
//end of combine averages procedure
//==================================
//continue the correctness procedure
//get the amendment for each zone-FORMSunit
foreach (array_keys($forms_to_correct_set) as $form) {
foreach ($zones as $zone) {
$dividend = 0;
foreach ($forms_to_correct_set[$form] as $asis) {
$summand = $correct_factors[$zone][$form]['factors'][$asis] * $combined_result[$zone][$asis]['average'];
$dividend += $summand;
}
$correct_factors[$zone][$form]['amendment'] = -1.0 * round($dividend / $correct_factors[$zone][$form]['divider'], 0, PHP_ROUND_HALF_UP);
}
}
//correct the averages in combined result
foreach ($correct_factors as $zone => $form_array) {
foreach ($form_array as $form => $factors_array) {
$combined_result[$zone][$form]['average'] += $factors_array['amendment'];
if ($combined_result[$zone][$form]['average'] < 0) {
$combined_result[$zone][$form]['average'] = 0;
}
}
}
//start the bodies count procedure
//================================
$bodies = [];
foreach ($entries_sums as $unit => $query_sums) {
$bodies[$unit] = [];
foreach ($query_sums as $query => $sums) {
$bodies[$unit] = array_merge($bodies[$unit], $sums);
}
$bodies[$unit] = $counter->getAverage($bodies[$unit]);
// $body_amendment = 0;
foreach ($zones as $zone) {
if ($zone != 'title') {
// $bodies[$unit] = $bodies[$unit];
$bodies[$unit] = $bodies[$unit] - $combined_result[$zone][$unit]['average'];
}
}
//get the average amendment from all zones
foreach ($correct_factors as $zone => $form_array) {
if (isset($correct_factors[$zone][$unit]) && $zone != 'title') {
$body_amendment = isset($body_amendment) ? $body_amendment += $correct_factors[$zone][$unit]['amendment'] : $correct_factors[$zone][$unit]['amendment'];
}
}
// $body_amendment = $body_amendment / 4.0;
if (isset($body_amendment)) {
$body_amendment = round($body_amendment, 0, PHP_ROUND_HALF_UP);
//average for 4 zones w/o title
$bodies[$unit] += $body_amendment;
if ($bodies[$unit] < 0) {
$bodies[$unit] = 0;
}
}
//correct the body amount with average amendment
}
return [$combined_result, $bodies];
}
public function substract($rest, $unit)
{
$lemmatizator = new Lemmatizator();
$norm_rest = str_replace(" - FORMS", "", $lemmatizator->makeUnitNormal($rest));
$norm_unit = str_replace(" - FORMS", "", $lemmatizator->makeUnitNormal($unit));
$rest_arr_normals = array_diff(explode(" ", $norm_rest), explode(" ", $norm_unit));
$rest_arr = array_combine(explode(" ", $norm_rest), explode(" ", $rest));
$new_rest = "";
foreach ($rest_arr_normals as $norm) {
$new_rest .= " " . $rest_arr[$norm];
}
return $new_rest;
}
